yon

561ue,avis,

Received 6 July 2012 the key factors inuencing the avour quality of fruits. The effect of ripening on chemical composition,physical parameters and sensory perception of three muskmelon (Cucumis melo L. reticulatus group) cul-tivars was evaluated. Signicant correlations emerging from this extensive data set are discussed in the

at harvest has also been shown to have a large impact on the sugarcontent (related to sweetness), volatile content (related to avourand aroma) and texture of melon fruit (Beaulieu & Grimm, 2001;Beaulieu, Ingram, Lea, & Bett-Garber, 2004; Beaulieu & Lancaster,2007; Pratt, 1971).

Harvesting rmer and early mature fruits is a commercialpractice commonly adopted in order to maximise post-harvest life

sometimes even before the clear development of an abscissionzone.

Due to the interactions of many parameters (e.g., sugar content,aroma prole, colour, texture) in determining fruit sensory charac-teristics, measuring a single composition parameter such as sugarcontent is seldom sufcient to reect an objective assessment ofoverall fruit avour quality. From an applicative perspective, acomprehensive assessment of avour quality is often unfeasibledue to the requirement of expensive analytical instrumentation,highly trained personnel and time- and labour-consuming proce-

Corresponding author. Tel.: +1 530 752 4374; fax: +1 530 752 8502.

Food Chemistry 139 (2013) 171183

Contents lists available at

he

lseE-mail address: fnegre@ucdavis.edu (F. Zakharov).Netted muskmelon (Cucumis melo L., reticulatus group), alsocommonly called cantaloupe, is an orange-eshed, sweet and aro-matic melon that is highly popular in the United States, represent-ing a large share of the produce market. In 2010, the estimated netdomestic use of muskmelon totaled over 2.6 billion pounds, andmuskmelon ranked fourth in U.S. annual per capita consumptionof fresh fruit after bananas, watermelons and apples (USDA-ERS,2010). Consumer surveys assessing overall preference for severalmuskmelon cultivars highlighted that avour, sweetness and tex-ture were important factors in determining consumer liking ofmelons (Lester, 2006). While these attributes are dictated by thespecic cultivar, or genetic makeup, of the muskmelon, maturity

2008). However, this practice is detrimental for avour quality be-cause it does not allow full development of the fruit aroma prole(Beaulieu, 2006; Beaulieu et al., 2004; Wyllie, Leach, & Wang,1996).

Typically, muskmelon fruit maturity in the eld is determinedby the extent of the development of an abscission layer (also calledslip in the trade) between the vine and the fruit. In California,melons are generally harvested at 3=4- to full-slip stage for localmarket distribution. However, genetic, environmental and agro-nomic factors often complicate maturity assessment by inuencingfruit physiology and the development of this abscission zone,resulting in variable postharvest fruit quality. In addition, melonsdestined for long distance transport are typically harvested earlier,Received in revised form 20 December 2012Accepted 28 December 2012Available online 7 January 2013

Keywords:Melon maturityRipeningAromaFlavourVolatilesSensory descriptive analysiszNoseHeadspace sorptive extraction

1. Introduction0308-8146/$ - see front matter 2013 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.foodchem.2012.12.042context of identifying potential targets for melon sensory quality improvement. A portable ultra-fast gas-chromatograph coupled with a surface acoustic wave sensor (UFGCSAW) was also used to monitoraroma volatile concentrations during fruit ripening and evaluated for its ability to predict the sensoryperception of melon avour. UFGCSAW analysis allowed the discrimination of melon maturity stagebased on six measured peaks, whose abundance was positively correlated to maturity-specic sensoryattributes. Our ndings suggest that this technology shows promise for future applications in rapid a-vour quality evaluation.

2013 Elsevier Ltd. All rights reserved.

during handling, shipping and storage of climacteric fruits (Kader,Article history: Numerous and diverse physiological changes occur during fruit ripening and maturity at harvest is one ofAn integrated approach for avour qualitmelo L. reticulatus group) during ripening

Simona Vallone a, Hanne Sivertsen b, Gordon E. AnthSusan E. Ebeler c, Florence Zakharov a,aDepartment of Plant Sciences, University of California, One Shields Avenue, Davis, CA 9bDepartment of Food Science and Technology, University of California, One Shields AvencDepartment of Viticulture and Enology, University of California, One Shields Avenue, D

a r t i c l e i n f o a b s t r a c t

Food C

journal homepage: www.ell rights reserved.evaluation in muskmelon (Cucumis

b, Diane M. Barrett b, Elizabeth J. Mitchama,

6, USADavis, CA 95616, USACA 95616, USA

SciVerse ScienceDirect

mistry

vier .com/locate / foodchem

ware program CSA, Compusense Five (Version 5.0, Compusense,

2 2 2 cm esh cubes obtained from the equatorial region of

misdures. While the availability of rapid methods for the detection ofexternal, visual quality has allowed the distribution of aestheticallysuperior fruit, the lack of rapid methods for avour quality controlmay hinder the delivery of more avourful fruit to consumers.

In our study we rst evaluated the effect of ripening on sensoryperception, chemical composition and physical measurements ofmelon fruit. Changes in chemical composition and physical proper-ties during the ripening process were then correlated to sensoryattributes in an attempt to predict avour perception by chemicalcomposition analysis. Finally, an ultra-fast gas-chromatograph wasevaluated for its ability to monitor changes in melon volatile con-centrations during ripening and to predict the sensory perceptionof avour.

2. Materials and methods

2.1. Plant material and sample preparation

Muskmelons (C. melo L., reticulatus group) cv. Navigator, MasRico and Thunderbird, were grown in Davis, CA (38.55 N,121.74 W), and provided by HM. Clause Seed Company (Modesto,CA, USA). Each of the three cultivars was planted on three differentdates with approximately 2-week intervals during the spring of2010 and grown on raised beds using standard commercial cultiva-tion practices with drip irrigation. Each cultivar plot correspondingto one planting date was used to supply one of three iterations offruit materials for sensory testing and physiochemical analysis.

Fruit were harvested at ve different maturity stages, rangingfrom early mature to fully ripe, between July and September2010. The maturity at harvest was assessed in the eld by examin-ing the presence or absence of the abscission zone formed aroundthe peduncle, commonly called slip. Early mature melonscorresponded to fruits that had reached full size, but did not showa visible abscission zone around the peduncle (thereafter namedpre-slip fruit). Fruits were classied as full slip when a fullydeveloped abscission zone (as evidenced by the development of acrack around the peduncle) was visible at harvest time. Underour growing conditions, the abscission zone developed veryquickly (within less than a day), making the standard 1=4-, - and3=4-slip maturity assessment impractical. Therefore, full-slip mel-ons were grouped in three classes, slip A (Sa), slip B (Sb) andslip C (Sc), based on increasing force needed to detach the fruitfrom the plant (force applied to detach the fruit: Sa > Sb > Sc). Sixmelons were chosen for each maturity stage based on the absenceof external and internal defects, and size homogeneity. Fruits wererinsed with tap water in order to remove dirt and dust, cut longi-tudinally into four wedges, and a further classication of thepre-slip fruit was performed visually based on the degree of col-our lightness of the esh: pre-slip light orange (PL) and pre-slipdark orange (PD).

Seeds and cavity tissue were removed, and two oppositewedges per fruit were selected for physiochemical analysis andthe other two opposite wedges were used for sensory analysis onthe same day.

2.2. Environmental conditions

Average and maximum daily air temperature and total solarradiation data were retrieved from the web-site of California Infor-mation Management System (CIMIS; http://www.cimis.water.ca.-gov) weather station, located 5 miles North-West (38.54 N,121.78 W) of HM. Clause Seed Company elds. Total solar radiationwas converted into photosynthetic active radiation (PAR) in 400

172 S. Vallone et al. / Food Che700 nm wavebands, according to Thimijan and Heins (1983), andexpressed in lmol m2 s1.the fruit. Hue angle, h, was calculated as h = arctan (a /b ).Puncture and compression assessments were performed to

evaluate fruit esh rmness, using a TA.XT2 Texture Analyzer(Texture Technologies, Scarsdale, NY, USA). Puncture testing wasperformed with a 100 g force load on the side of a 2 2 2 cmesh cube obtained from the equatorial region of the fruit. A5 mm diameter at-head stainless steel cylindrical probe travelledat a rate of 1 mm s1 for a total of 6 mm. The area under the curvefrom 0 to 6 mm was used as the puncture measurement. Compres-sion testing using a 38 mm at compression probe was performedwith a 100 g force load on the side of a 2 2 2 cm cube. Pretestspeed was 10 mm s1 with a test speed of 0.5 mm s1, followed bya post-test speed of 10 mm s1. The compression measurementdetermined from the graph was total force area (N s). Six melonGuelph, Ontario, Canada, 2008), in individual booths and undernormal uorescent white light, at ambient room temperature(20 C). Water and unsalted crackers were used as rinsing agentsbetween samples.

A modied quantitative descriptive analysis (QDA) (Stone, Sidel,Oliver, Woolsey, & Singleton, 1974) was used to evaluate the sam-ples. Five samples per cultivar, representing all ve maturitystages, were evaluated during each session and each cultivar eval-uation was repeated three times (corresponding to the three differ-ent planting dates). Panellists evaluated the samples one by one, ona computer screen, according to the order presented in the soft-ware program. Appearance of the fruit was evaluated rst basedon the overall average colour intensity and unevenness of colourfor the ve melon balls. Panellists then smelled the sample, andevaluated the aroma attributes. Two sample balls were used forevaluation of the texture attributes. Panellists swallowed the sam-ples after completing the evaluation. Flavours, tastes and avourlasting sensation were evaluated on the remaining three sampleballs.

2.4. Physical and chemical analysis

2.4.1. Physical measurementsColour measurement was performed using a Minolta Colorime-

ter (CR-300, Minolta, Ramsey, NJ, USA). L, a, b values were re-corded on six replicates per sample, from the side of2.3. Sensory analysis

2.3.1. Panellist recruiting and trainingStudents and staff with various backgrounds in sensory testing

were recruited from the UC Davis campus to participate in sensorypanel training.

Nine panellists underwent six one-hour training sessions, over aperiod of 3 weeks. During the initial training sessions, panellistsgenerated and agreed upon a list of sensory attributes presentedin Table 1. The attributes were rated on a 10 cm unstructured scale,anchored at the ends with none and strong, except for colourintensity, which ranged from pale orange to strong orange,and unevenness of colour, which ranged from even to uneven.

2.3.2. Sensory descriptive analysisMelon balls (1.3 cm in diameter) carved from melons of the

same cultivar and maturity stage were gently mixed in a mixingbowl, and a set of ve balls was put into ve 162-mL plastic cupswith lids labelled with random three digit codes. The samples wereserved in complete randomised order, as established by the soft-

try 139 (2013) 171183cubes were analysed for puncture and compression for each culti-var, maturity level, and planting date.

ve

hite

l aroity a

suga

ngin

thba

tron

with

teet

emisTable 1Sensory attributes and reference standards.

Attribute Attribute descriptions

AppearanceColour intensity The intensity of the most dominant colour of the sample (all

orange to strong orangeUnevenness of

colourThe degree of evenness of the colour, ranging from even (no wthe sample

AromaOverall aroma The intensity of aroma of any type, ranging from weak overalFruity aroma The intensity of fruity aroma like mixed fruit juice, a fresh fru

Marshmallowaroma

The intensity of sweet aroma, includes the sweet sensation ofnone to strong sweet aroma.

Cucumberaroma

The intensity of green, fresh aroma, like cucumber, squash, ra

Musky aroma The intensity of the spicy aroma of cedar wood, like cedar mofrom none to a strong musky aroma

Buttery aroma The intensity of the aroma of butter, ranging from none to a s

TextureJuiciness The amount of juice released in the sample when biting into it

to lots of juice releasedFirmness The force required to compress the sample between the back

S. Vallone et al. / Food Ch2.4.2. Chemical measurementsSoluble solids content (SSC) was determined by refractometry

(Reichert AR6 Series, Depew, NY, USA) using 200 lL of juice ob-tained from squeezing melon balls (using a garlic press) sampledfrom blossom end, stem end and equatorial regions of the melons.Titratable acidity (TA) was determined by diluting 4 g of juice (ob-tained as above) in 20 mL deionised water and titrating to pH 8.2using an automatic titrator (Radiometer TitraLab Tim850, Radiom-eter Analytical SAS, Lyon, France). TA is reported as citric acidequivalents. Each measurement was performed in triplicate foreach sample. For ethylene and carbon dioxide analysis, 25 g of mel-on pieces (obtained randomly along melon wedges using an11 mm i.d. cork borer) were enclosed in air-tight glass vessels(250 mL volume). After equilibration for 15 min, a 10-mL head-space sample was analysed by gas chromatography coupled withame ionisation detector using a Varian 3800 (Walnut Creek, CA,USA) equipped with a 6 m 3 mm alumina column held at 50 C.For carbon dioxide analysis, 10-mL headspace sample was injectedinto a rapid gas analyzer (VIA510; Horiba, Fukuoka, Japan). Ethyl-ene and CO2 concentrations were calculated by comparing thesample response to authentic standard curves.

Juice samples obtained as described for SSC and TA measure-ments were used for sugar and acid analysis. Fructose, glucose, su-crose, citric, malic and glutamic acid concentrations wereevaluated by enzymatic assays as described in Vermeir, Nicolai,Jans, Maes, and Lammertyn (2007) using enzyme reagent kits

Crunchiness The amount of sound generated when chewing the back teeth, rancrunchy sound

Fibrousness The presence of bres in the esh, ranging from no bres to manyTaste and avourSweet taste The intensity of sweet taste like sugars, but also like candy (marsh

strong sweet tasteSour taste The intensity of fresh sour taste like Granny Smith apples, ranging

Fruity avour The intensity of the avour of fruit juice of mixed fruits, ranging f

Bitter taste The intensity of bitter taste, like caffeine, but also like the peel of cua strong bitter taste

After avour How long the avour lasts in the mouth, ranging from a few seconReferences

melon balls included), ranging from very pale None

spots in esh), to uneven (many white spots) in None

ma, to strong aroma present in the sample Noneroma, ranging from none to strong fruity aroma Dole mixed fruit juice

(orange, peach, mango juice)Scale point: 10

r, marshmallows, and bubblegum, ranging from Marshmallows, small sizeScale point: 10

g from none to strong, green aroma English cucumber, peeled andcut into 1 cm slicesScale point 10

lls, but also the musky smell of animal, ranging A piece of cedar woodScale point: 10

g buttery aroma Real butter, ca. 1 cm3

Scale point: 10

the front teeth, ranging from no juice released, Ripe peachScale point: 10

h, ranging from soft to rm Banana, yellow, ripe

try 139 (2013) 171183 173(R-Biopharm, Marshall, MI, USA) according to the manufacturersinstructions. Standard solutions were used to verify the accuracyof the method.

2.4.3. Volatile compound analysis2.4.3.1. Sample preparation. For a detailed sample preparation, seeVallone, Lloyd, Ebeler, and Zakharov (2012). Briey, after removingthe seeds and seed cavity tissue, melon balls (2.5 cm in diameter)from six fruits were pooled and mixed, and then 200 g of samplewere homogenised with 200 mL of saturated CaCl2 solution and50 lL of a 100 mM solution of 2-methylbutyl isovalerate in meth-anol, added as an internal standard. Then, 5-mL aliquots of samplewithout foam were pipetted into 20-mL glass amber vials. Thevials, sealed with a steel screw cap tted with a Teon/silicon sep-tum, were ash frozen in liquid nitrogen, and stored at 80 C untilanalysis. For volatile analysis, samples were thawed for one hour atroom temperature immediately prior to analysis.

2.4.3.2. Headspace sorptive extraction and gas chromatography massspectrometry. Headspace sorptive extraction (HSSE) and gas chro-matography mass spectrometry (GCMS) was performed usingan Agilent (Santa Clara, CA, USA) gas chromatograph (7890A) andmass spectrometer detector (5975C), equipped with a thermaldesorption unit (TDU) and cryo-cooled injection system (CIS) (Ger-stel Inc., Mlheim an der Ruhr, Germany). The headspace samplingwas performed with a 10-mm magnetic stir bar (also called

Scale point: 01English cucumber with peelScale point: 10

ging from no sound, like a banana, to a lasting Granny Smith apple wedgeScale point: 10

bres in the sample No reference

mallows or bubblegum), ranging from none to a Marshmallow, small sizeScale point: 10

from none to a strong sour taste Granny smith appleScale point: 10

rom none to a strong fruity avour Dole mixed juice (orange,peach, mango):Scale point: 10

cumber, or an unripe fruit, ranging from none to English cucumber peelScale point: 10

ds to longer than 20 s None

(data not shown). In some muskmelon varieties, a seasonal varia-

ma, but was scored high for cucumber aroma, rmness and

misTwister) coated with 24 lL of polydimethylsiloxane (Gerstel Inc.,Germany). The bar was suspended above the sample in the 20-mLamber vial and exposed to the headspace for 30 min at room tem-perature, while the sample was stirred at 550 rpm. For thermaldesorption, the following parameters were used: initial TDU tem-perature of 30 C, heated to 40 C at 720 C min1, then to 250 Cat 60 C min1 and held for 5 min; solvent venting as desorptionmode and vent time of 0.01 min. Simultaneously, the analyteswere transferred into the PTV-injector where they were cryogeni-cally focused at 80 C, using liquid nitrogen. The injection modewas set as solvent vent; vent ow of 50 mL min1 and vent pres-sure of 0 psi until min 0; purge ow to split vent of 5 mL min1

at min 1; gas saver of 20 mL min1 after 2 min. The initial injectiontemperature was programmed at 80 C, then heated to 280 C at12 C s1 and held for 5 min. The GC was equipped with a DB-5MScapillary column (30 m 0.25 mm; lm thickness 0.25 lm). Theinjector temperature was 220 C and helium carrier gas ow ratewas 1.2 mL min1. MS parameters were as follows: the transfer-line was set to 230 C; the source was 230 C; the quadrupole tem-perature was 150 C; and the scan range was from 30 to 300m/z ata rate of 3 scans s1. Spectral deconvolution was performed withAMDIS software version 2.69 (Stein, 1999) and spectral alignmentwas performed with Mass Proler Professional (version B2.01; Agi-lent, Santa Clara, CA, USA).

Volatile compound identity was conrmed by matching reten-tion time and mass spectra of authentic standards whenever pos-sible. The percent relative response ratio of each volatilecompound was calculated using the internal standard peak area.

All samples were analysed in triplicate.

2.4.3.3. Ultra-fast chromatography and surface acoustic wave sensoranalysis. An ultra-fast gas chromatograph (GC) coupled with a sur-face acoustic wave (SAW) sensor called zNose (model 4500; Elec-tronic Sensor Technology-EST; Newbury Park, CA, USA), was usedas a rapid method for volatile compounds analysis. The UFGCSAW analysis was carried out in two steps, sampling and injection.A detailed description of the sampling and injection parameters isfound in Vallone et al. (2012).

2.5. Statistical analysis

The sensory data was captured by the Compusense softwareprogram, which converted each panellist rating to a value between1.00 and 9.00 (with 1.00 anchored at the low intensity end and9.00 anchored at the high intensity end of the scale). A four-wayanalysis of variance (generalised Linear Model) was run with amixed model (9 judges 5 ripening stages 3 cultivars 3 plant-ing dates). The judges were evaluated as a random effect. The over-all panel performance, judge repeatability and discriminationability were evaluated on a randomly selected set of replicatesusing the software PanelCheck version 1.4.0 (Tomic et al., 2010).

A three-way analysis of variance (ANOVA) was performed toanalyse the effects of maturity at harvest, cultivar and plantingdate on sensory attributes, physical and chemical measurements.Tukeys test was performed for all pair-wise comparisons to evalu-ate statistically signicant differences among the means. A princi-pal component analysis (PCA) was performed to reduce thedimensionality of the data, and to visualise relations among allthe variables (sensory and physiochemical). First, Pearsons corre-lation coefcients were calculated for each pair-wise combination,using log transformed and mean centered data. Then the new vari-ables (principal components) were obtained from the eigenvectorsof the estimated correlation matrix of the original variables (Bor-

174 S. Vallone et al. / Food Chegognone, Bussi, & Hough, 2001). Since the variables measuredhad different units, the correlation matrix was selected to stan-dardise the original variables and obtain a variance equal to 1.crunchiness. The Thunderbird cultivar was the most bitter andhad the lowest sweet taste and fruity avour intensities. Previ-ous studies comparing sensory attributes in different melon culti-vars also found signicant differences in taste, aroma and textureperception (Senesi, Di Cesare, Prinzivalli, & Lo Scalzo, 2005; Senesi,Lo Scalzo, Prinzivalli, & Testoni, 2002).

The different cultivars tended to follow similar changes in sen-sory perception during maturation. The sensory scores for colourintensity, overall aroma, fruity aroma, marshmallow aroma,musky aroma, juiciness, sweet taste, fruity avour andafter avour increased with increasing maturity, while sensoryscores for cucumber aroma, rmness, crunchiness, brous-tion in colour, SSC and sugar content has been observed (Beaulieu,2005; Beaulieu, Lea, Eggleston, & Peralta-Inga, 2003).

3.1.1. Sensory evaluationJudge performance quality, based on the evaluation reproduc-

ibility on a replicate set of samples, was evaluated using Panel-Check (Tomic et al., 2010). This analysis showed that the panelused the attributes in a similar and consistent way (data notshown).

Signicant differences in most sensory attributes were foundacross maturity stages and among the three cultivars (Table 2). Sig-nicant planting date and interactions effects were also observedfor various sensory attributes.

When scores for each sensory attribute were averaged across allmaturity stages and planting dates, the cultivar Navigator was gen-erally perceived to have the highest intensity of colour, overallaroma, sweet taste, and fruity avour, and a lower intensityof cucumber aroma and bitter taste (Table S1). Mas Rico gener-ally had the lowest perceived intensity of colour and fruity aro-The Scree test was used to select the number of principal compo-nents to retain. Regression analysis was performed on selectedvariables for which the assumption of dependence was satised,and the coefcient of determination and signicance of the slopewere calculated. The analysis of variance and Tukeys test wereperformed using the statistical program SAS (version 9.1.3; SASInstitute Inc., Cary, NC, USA), while JMP (version 10.0.0; SAS Insti-tute Inc., Cary, NC, USA) was used for principal component andregression analysis. Pearsons correlation analysis was performedusing the package Hmisc (Harrell & with contributions from manyother users, 2012) in the statistical programming language R (R2.14.0-Development Core Team 2008, Vienna, Austria).

3. Results and discussion

3.1. Physiochemical and sensory evaluation of three melon cultivars atdifferent maturity stages

Maturity at harvest is one of the key factors inuencing melonquality, due to the numerous physiological changes that take placeduring the ripening process. As expected with eld-grown fruit,signicant differences were observed in some sensory and physio-chemical parameters, not only as a function of maturity and culti-var, but also as inuenced by environmental factors (i.e., plantingdate) (Table 2, Table S3). Changes in natural environment that oc-curred between the three growing periods may have inuenced thecomposition of the fruits at harvest time. An increase in maximumair temperature as well as a decrease in photosynthetic active radi-ation (PAR) were observed between the rst and last planting date

try 139 (2013) 171183ness and sour taste attributes generally decreased as maturityincreased. The largest increase in perceived intensity was foundbetween pre-slip light and dark stages for colour intensity,

Cu

0.00.60.60.60.30.20.50.10.00.00.00.30.40.10.00.00.1

g da

emissweet taste and fruity avour attributes. Large changes in thetexture attributes, rmness and crunchiness, were observedbetween pre-slip dark, slip A and slip B stages. A larger in-crease in sweet taste and fruity avour attributes was per-ceived in Mas Rico from the slip A to slip B stages, comparedto the other two varieties. Musky aroma perception increasedsteadily and to a greater extent in Navigator throughout ripening.

Our results are in general agreement with similar studies eval-uating sensory attributes of melon at different ripening stages,where the largest differences in rmness, taste and avour attri-butes were perceived between the unripe and ripe stages(which may correspond to pre-slip and slip A/B stages in ourstudy), and little difference in these attributes was noted betweenripe and overripe stages (similar to slip B and slip C in ourstudy) (Senesi et al., 2005). Similarly, Beaulieu et al. (2004) ob-served signicant changes in sensory attributes such as sweetaromatic, sweet taste, surface wetness and hardness ofmuskmelons harvested at stage 1/4 slip or 1/2 slip, but not inlater maturity stages (3/4 slip and full slip). These observations

Table 2Analysis of variance: main effects and signicance levels for sensory attributesa

Attribute Planting date Cultivar Stage

Colour intensity 0.1163 0.0474

Table 3Physical and chemicala parameters measured in three melon cultivars harvested at ve maturity stages from three planting dates.

Cultivarb Maturityc Planting date Puncture Compression L a b Hue Ethylene CO2 SSC TA Glutamic acid Citric acid Malic acid Sucrose Glucose Fructose(N) (N) (lL kg1 h1) (mL kg1 h1) (%) (%) (g L1) (g L1) (g L1) (g L1) (g L1) (g L1)

T PL 1 18.0 40.2 65.1 11.6 37.0 0. 30 2.7 186.7 8.7 0.08 0.29 3.06 0.34 18.0 28.0 29.02 19.1 36.9 63.5 9.1 34.2 0.26 1.7 102.4 7.8 0.31 0.17 2.98 0.32 14.6 26.7 28.63 18.7 50.6 64.8 7.9 32.6 0.24 6.1 170.7 8.7 0.06 0.22 3.17 0.31 12.1 25.8 25.8

M PL 1 19.0 46.0 66.2 11.0 37.3 0.29 3.2 193.4 10.7 0.27 0.52 2.96 0.45 37.3 26.1 26.52 18.0 45.7 66.3 9.7 34.7 0.27 3.6 201.8 8.0 0.07 0.21 2.48 0.43 16.3 26.2 25.53 21.8 48.2 66.9 9.8 35.2 0.27 3.4 127.9 6.6 0.08 0.29 2.78 0.38 8.4 24.1 24.2

N PL 1 16.5 39.9 67.4 10.5 35.5 0.29 3.8 236.0 9.4 0.04 0.73 2.46 0.41 23.3 28.2 28.62 16.7 38.7 66.3 9.9 32.8 0.29 13.9 189.6 7.9 0.95 0.27 2.56 0.34 15.3 26.2 27.03 16.2 41.4 67.9 10.5 35.4 0.29 6.0 249.8 9.0 0.03 0.40 2.68 0.02 14.7 21.8 27.2

T PD 1 12.6 27.4 65.9 10.7 38.0 0.27 3.3 177.8 9.3 0.06 0.64 2.41 0.40 26.3 29.1 31.22 11.8 23.3 63.9 10.2 35.0 0.28 8.3 126.8 9.9 0.27 0.55 2.54 0.36 30.7 25.9 27.63 18.6 36.8 62.6 10.3 35.0 0.29 2.9 258.3 10.0 0.12 0.35 3.16 0.36 30.6 26.0 27.1

M PD 1 16.8 41.9 65.6 11.9 38.3 0.30 3.3 203.8 11.2 0.23 0.88 2.73 0.51 43.7 25.4 26.42 17.0 34.0 64.1 10.9 35.4 0.30 16.2 128.9 9.0 0.06 0.55 1.98 0.49 25.1 24.7 24.83 19.4 36.2 67.9 10.5 34.8 0.29 5.3 151.9 7.6 0.06 0.64 2.21 0.48 17.9 22.2 22.4

N PD 1 15.7 37.1 65.5 11.9 37.6 0.31 7.2 210.2 10.4 0.03 1.36 2.22 0.49 34.2 26.0 28.32 11.6 27.4 62.7 10.2 33.3 0.29 12.1 253.7 10.0 0.79 1.39 1.61 0.45 27.1 26.6 27.83 16.4 35.7 65.8 11.5 36.3 0.31 6.1 252.9 10.6 0.05 1.35 2.17 0.16 21.4 22.4 19.7

T Sa 1 6.6 14.6 68.6 9.9 35.6 0.27 21.9 240.1 8.5 0.06 0.68 1.94 0.24 21.7 20.8 23.62 7.7 15.8 58.1 10.1 33.0 0.30 13.9 70.2 8.8 0.32 0.84 2.15 0.30 23.4 24.6 28.63 7.6 17.5 63.4 10.0 35.2 0.28 19.0 175.7 8.3 0.07 0.57 2.87 0.23 23.8 25.8 29.8

M Sa 1 7.8 18.3 62.6 11.9 38.1 0.30 15.7 172.7 10.6 0.24 1.11 2.86 0.37 38.6 26.1 28.12 13.1 32.3 63.0 11.1 35.7 0.30 13.8 160.0 9.4 0.06 0.69 1.67 0.43 28.9 25.4 26.33 11.8 23.6 65.1 10.2 34.4 0.29 13.8 172.4 8.0 0.06 1.20 1.93 0.41 24.5 20.1 20.6

N Sa 1 8.8 21.8 64.5 11.7 36.7 0.31 19.1 175.8 10.2 0.04 1.49 2.33 0.42 33.4 27.0 28.82 11.2 25.5 65.3 11.7 36.4 0.31 20.5 295.3 9.1 0.68 1.26 0.85 0.49 29.1 23.3 25.33 9.5 20.5 66.1 11.8 36.7 0.31 20.3 263.6 9.6 0.05 1.13 2.17 0.02 16.1 22.1 27.3

T Sb 1 5.6 10.9 67.1 11.1 37.8 0.28 15.5 213.5 8.0 0.06 0.98 1.93 0.29 22.1 21.3 25.82 9.7 23.4 61.4 10.7 36.1 0.29 13.5 152.4 8.9 0.30 1.22 1.46 0.41 28.3 21.6 26.33 7.7 14.3 61.6 11.1 35.8 0.30 19.2 267.2 8.7 0.06 0.72 2.24 0.30 25.5 20.7 24.9

M Sb 1 11.7 31.6 62.8 11.3 37.2 0.29 23.9 263.5 10.3 0.24 1.23 2.47 0.44 40.6 22.6 24.82 7.7 14.3 56.2 10.8 33.6 0.31 11.3 138.9 9.6 0.06 0.60 2.18 0.44 36.1 21.0 23.53 9.2 20.8 65.2 11.0 35.5 0.30 19.1 173.0 9.1 0.06 0.96 2.07 0.40 34.3 19.8 22.2

N Sb 1 7.2 14.1 60.9 11.0 34.1 0.31 16.1 176.8 9.2 0.04 1.53 1.67 0.48 31.4 22.0 26.12 9.2 20.8 62.2 11.0 34.8 0.30 11.0 253.4 8.1 0.76 1.59 0.92 0.49 21.3 21.0 23.73 8.6 22.1 64.6 11.9 36.8 0.31 20.4 208.9 9.4 0.05 1.97 1.78 0.04 29.0 28.0 35.9

T Sc 1 4.1 11.6 67.6 10.8 37.1 0.28 16.4 201.8 8.2 0.06 0.83 1.77 0.31 27.7 18.9 26.22 6.1 11.6 62.6 10.2 35.2 0.28 12.6 59.7 8.7 0.29 1.24 1.88 0.38 25.7 21.8 26.23 6.1 16.6 61.4 9.6 33.6 0.28 20.4 212.7 8.7 0.06 0.56 1.91 0.33 32.6 18.3 24.3

M Sc 1 5.4 16.6 63.5 11.5 37.4 0.30 19.8 221.0 9.6 0.30 0.64 2.87 0.29 35.8 21.7 26.12 8.7 22.8 59.6 11.9 36.0 0.32 14.2 167.4 10.3 0.06 0.78 2.24 0.49 42.2 19.6 21.83 9.1 22.0 62.5 12.4 37.2 0.32 13.6 210.6 9.8 0.07 0.68 2.19 0.48 42.8 18.2 20.1

N Sc 1 6.5 16.8 61.7 11.1 35.8 0.30 15.8 175.6 9.0 0.04 1.73 1.62 0.55 31.8 18.3 24.42 6.4 15.8 59.4 10.9 34.5 0.31 7.1 201.0 8.6 1.21 0.85 1.29 0.53 28.5 18.7 24.03 8.1 18.0 64.4 12.1 37.1 0.31 19.2 217.0 9.0 0.05 1.70 0.79 0.02 5.4 19.0 21.5

Data are reported as mean value of three replicates.a Volatile compounds are reported in Table 4.b Cultivar: T = Thunderbird; M = Mas Rico; N = Navigator.c Maturity stage at harvest: PL = pre-slip light; PD = pre-slip dark; Sa = slip A; Sb = slip B; Sc = slip C.

176S.V

alloneet

al./FoodChem

istry139

(2013)171

183

and 0.75). Ethylene has been shown to play an important rolein the regulation of many ripening processes including fruit soften-

emiscal state during ripening was monitored using esh ethylene pro-duction rates and respiration rates. These parameters were highlyvariable and affected by environmental factors; however, a generaltrend in ethylene production could be noted, as there was asignicant and consistent sharp increase in ethylene productionrate between the pre-slip dark (7.2 lL kg1 h1) and slip A(17.6 lL kg1 h1) stages in all three cultivars. This sharpincrease in ethylene production may constitute a signal for theinitiation of faster ripening and may explain the sharper decreasein rmness between these two stages, as ethylene has beenshown to regulate cell wall breakdown in melon (Nishiyamaet al., 2007).

Comprehensive analysis of melon volatile proles was carriedout using headspace sorptive extraction coupled with GCMS(HSSE-GCMS), and 82 volatile compoundswere detected (Table 4).Consistent with previous reports (Beaulieu, 2006; Beaulieu &Grimm, 2001; Senesi et al., 2005; Wang et al., 1996; Wyllie, Leach,Wang, & Shewfelt, 1995), the total volatile content signicantly in-creased during ripening. Total volatile concentration in pre-slipand slip A fruit was less than 20% and 50%, respectively, that offully ripe fruits (slip B). Among all the volatiles detected,twenty-eight compounds were signicantly affected by fruit matu-rity at harvest (Table 4).

Volatile esters were the predominant constituents of melonesh aroma, representing from 48% to 94% of the total volatilesmeasured in pre-slip and slip B fruits, respectively.

Several volatiles measured in this study were previously identi-ed as important muskmelon aroma compounds (Beaulieu &Grimm, 2001; Homatidou, Karvouni, Dourtoglou, & Poulos, 1992;Jordan, Shaw, & Goodner, 2001; Kourkoutas, Elmore, & Mottram,2006; Wyllie & Leach, 1992; Wyllie, Leach, Wang, & Shewfelt,1994). Various esters quantied in our study were previouslyinvestigated for their sensory signicance in melon aroma andhave been identied as odour active compounds by gas chroma-tographyolfactometry (Jordan et al., 2001; Schieberle, Ofner, &Grosch, 1990; Wyllie et al., 1994). As reported by these authors,some esters measured in our study were found to activelycontribute to the fruity and oral notes characteristic of melon ar-oma, including isobutyl acetate (described as oral), ethyl 3-(methylthio)propanoate (clean/fresh/melon/green), heptyl acetate(clean/fresh/oral), and (Z)-3-hexen-1-yl acetate (green/herbal/banana).

The aldehyde fraction greatly decreased in ripe fruits. Althoughenvironmental conditions affected the concentrations of nearly allaldehydes, hexanal, 2-heptenal and 2,4-nonedienal were generallypresent at higher concentration in early mature fruits compared tofully ripe fruits. Beaulieu and Grimm (2001) also found higher con-centration of 2-heptenal in immature muskmelons (Cucumis meloreticulatus group cv. Sol Real and Athena). Hexanal and (E)-2-non-enal, both measured in our study, have been reported to be respon-sible for green and cucumber-like notes in melon and in cucumber(Cucumis sativus L.) aroma (Schieberle et al., 1990; Ullrich & Grosch,1987). The remaining fraction of the total volatile prole was com-prised of sulphur-containing esters, alcohols, ketones, terpenoids,norisoprenoids and unidentied compounds. The concentrationof sulphur-containing esters, alcohols and terpenoids signicantlyincreased as fruits ripened.

In addition to ripening-associated changes in melon aroma,quantitative differences in volatile concentrations were observedbetween the three cultivars investigated in this study. Cultivar-specic differences in aroma proles were also reported in previ-ous studies (Beaulieu & Grimm, 2001; Obando-Ulloa, Ruiz, Mon-forte, & Fernandez-Trujillo, 2010; Senesi et al., 2005; Wyllie &

S. Vallone et al. / Food ChLeach, 1992; Yamaguchi et al., 1977), highlighting a strong geneticcontrol of the aromatic trait.ing in melon (Ayub et al., 1996; Hadeld et al., 2000; Pech et al.,2008). For example, in ripening Charentais melons (Cucumis melo,cantalupensis group), fruit softening due to cell wall disassemblywas shown to be an ethylene-dependent mechanism (Nishiyamaet al., 2007). Ethylene has also been shown to be involved in regu-lating volatile production and aroma development during melonripening (Bauchot, Mottram, Dodson, & John, 1998; El-Sharkawyet al., 2005; Flores et al., 2002). Consistently, we observed that infully ripe fruits, sensory attributes such as overall aroma inten-sity, juiciness, fruity avour, marshmallow aroma and fru-ity aroma were signicantly correlated with ethylene productionOn average, the total volatile content was higher in Navigatorthan in Thunderbird and Mas Rico (68% and 36% of the total vola-tile content of Navigator, respectively).

Although sulphur-containing esters were measured in all threecultivars, a signicant increase in ethyl-(methylthio)acetate andethyl-3-(methylthio)propionate concentrations was observed dur-ing ripening only in Navigator. In a broad screening conducted on26 melon cultivars with the intent to investigate the incidence ofsulphur-containing esters in the aroma prole, ethyl-(methyl-thio)acetate was frequently detected and its presence in the aromaextract appeared to be cultivar dependent (Wyllie & Leach, 1992).

3.2. Investigating relationships between sensory attributes andphysiochemical parameters

To investigate relationships among sensory perception andphysiochemical parameters, and the effects of ripening on overallmelon avour, a correlation analysis was performed on all vari-ables, followed by principal component analysis (PCA) on the cor-relations matrix to visualise the relationships. The rst sixcomponents explained 70.3% of the total variation, with the rsttwo components accounting for 43.8% and 10.5% of the total vari-ation (Fig. 1). While variation along the second component waslikely due to environmental effects, on the rst component, separa-tion of the samples was primarily driven by maturity at harvest.Early mature fruits (pre-slip light and dark) clustered on the leftside of the plot and were negatively correlated with the more ma-ture fruits (slip B and slip C), which were grouped on the oppo-site quadrant (Fig. 1).

Indeed, all the variables that had signicantly higher values inpre-slip fruits were found on the left quadrant of the plot. Fleshtexture (compression and puncture) was positively correlatedwith sensory rmness (r = 0.87 and 0.92) and crunchiness(r = 0.88 and 0.92) sensory attributes, which were scored high inpre-slip fruits (Fig. 1). Cucumber aroma was signicantly corre-lated to rmness, as measured with both Texture Analyzer andby the sensory panel (rP 0.81). Cucumber aroma was also corre-lated with the volatile compound, 2-heptenal (r = 0.78), which isderived from oxidation/degradation of fatty acids and has beenpreviously reported to contribute to the avour of fresh cucumbers(Grosch & Schwarz, 1971; Ullrich & Grosch, 1987). Signicant asso-ciations between the sensory attributes hardness, cucurbit(which could correspond to our sensory rmness and cucumberaroma) and total aldehyde levels were also found in the WesternShipper cultivar Sol Real (Beaulieu & Lancaster, 2007).

Ethylene was produced at higher levels in fully ripe fruits andwas also negatively correlated to rmness evaluated using bothinstrumental (compression and puncture: r = 0.69 and 0.74)and sensory assessments (crunchiness and rmness: r = 0.74

try 139 (2013) 171183 177(rP 0.59). In addition, ethylene production was highly correlatedwith the concentrations of numerous volatile compounds, particu-

Table 4Analysis of variance of volatile compounds: signicance levels of main effects and interactions.

Volatile compound CAS # RTa RIb RIc Cultivar Maturity Planting date Maturity cultivar Maturity planting date Cultivar planting date1 Ethyl acetate 141-78-6 2.12 674 605/628d 0.045 0.005 0.735 0.532 0.827 0.3082 Propyl acetate 109-60-4 2.99 722 707/720d 0.023 0.015 0.167 0.087 0.249 0.1883 Ethyl isobutyrate 97-62-1 3.68 761 751/762d

53 Methyl octanoate 111-11-5 14.95 1125 1126d

mis180 S. Vallone et al. / Food Chelarly esters, e.g., 2-methylbutyl acetate (r = 0.73), pentyl acetate(r = 0.66), heptyl acetate (r = 0.63), and hexyl acetate (r = 0.63).

The fruity aroma sensory attribute was signicantly and pos-itively correlated with many volatiles, including benzyl acetate,pentyl acetate, hexyl acetate, benzaldehyde, eucalyptol, heptyl ace-tate, isobutyl butanoate and prenyl acetate (correlation coefcientsrP 0.76). Benzyl acetate, hexyl acetate and eucalyptol have previ-ously been described in various melon types as characteristic im-pact avour or aroma compounds (CIFAC) with fruity, oral and/or fresh odour characters (Beaulieu, 2005; Jordan et al., 2001;Kourkoutas et al., 2006).

Two isomers of 2,3-butanediol diacetate detected in our sam-ples have also been reported in several melon varieties (Beaulieu& Grimm, 2001; Homatidou et al., 1992; Kourkoutas et al., 2006;Wyllie & Leach, 1990). The racemic mixture was described to havea sweet smell and high odour threshold, and therefore suggestednot to signicantly impact the aroma of melon cv. Golden Crispy(Wyllie & Leach, 1990). In our samples, however, both isomerswere positively correlated to the marshmallow aroma sensoryattribute (r = 0.76).

Musky aroma was positively correlated to the presence ofethyl-(methylthio)acetate and ethyl butanoate (rP 0.80). In ourstudy the perception of musky aroma was higher in Navigator;the volatile prole for this varietywas also signicantly richer in sul-phur-containing esters compared to the other varieties. It has beenreported that trace amounts of some sulphur compounds, such asethyl-(methylthio)acetate, might contribute to the musky note inmuskmelon (Cucumis melo cv. Makdimon) (Wyllie et al., 1994).

Fig. 1. Biplot from Principal Component Analysis of physical and chemical paramsquare = Thunderbird; triangle = Mas Rico; circle = Navigator. Maturity is represented blight grey = pre-slip dark; lled symbols, dark grey = slip A; lled symbols, black = slip B;Table 4; only the volatiles signicantly inuenced by maturity at harvest, cultivar and/orfrom the origin of the axis to a particular parameter indicates how well the variation intry 139 (2013) 171183Sweet taste was positively correlated to concentrations of su-crose (r = 0.53), as expected, and more surprisingly, to glutamicacid (r = 0.53). Human sweet and umami taste receptors, locatedin taste buds, respond to diverse sweetening components (suchas sugars, synthetic sweeteners and sweet-tasting proteins) andglutamate and purine nucleotides (such as 50-inosine monophos-phate and 50-guanosine monophosphate), respectively. Interest-ingly, sweet and umami receptor protein complexes have beenshown to share a common subunit (Li et al., 2002), and severalstudies in rats have suggested that glutamate can activate bothtypes of receptors, likely evoking both sweet and umami tastes(Heyer, Taylor-Burds, Tran, & Delay, 2003; Ninomiya et al., 2000;Sako & Yamamoto, 1999; Yamamoto et al., 1991). While the roleof glutamate as avour enhancer is well established in human tasteperception, most studies have focused on the effect of glutamateconcentration on taste modalities (e.g. saltiness, bitterness andumami) associated with savoury foods (Fuke & Ueda, 1996; Yeo-mans, Gould, Mobini, & Prescott, 2008). On the other hand, the spe-cic effect if any of glutamate on sweetness perception inhumans is still unclear (Kemp & Beauchamp, 1994; Maga, 1983).Our results suggest an intriguing role for glutamate in sweetnessperception in melons that warrants further investigation.

3.3. Evaluation of melon avour quality using a portable ultra-fast gaschromatograph (the zNose)

An analysis of variance performed on a total of 22 peaks mea-sured by UFGCSAW showed that the abundance of 6 peaks (peak

eters and sensory attributes. Symbol shape corresponds to different cultivars:y different shades of grey: open symbols, light grey = pre-slip light; lled symbols,open symbols, black = slip C. Numbers correspond to identied volatiles reported inplanting date effects were included in the analysis. The length of the dotted arrowsthat parameter is explained by the model.

2, 5, 9, 13, 14 and 17) was signicantly inuenced by maturity atharvest regardless of the cultivar, although some quantitative dif-ferences were also observed between cultivars (Table S3).

To our knowledge, no previous UFGCSAW studies investigatedrelationships between peaks detected and sensory attributes. To-wards this goal, we selected peaks and sensory attributes that weresignicantly affected by maturity at harvest, cultivar and/or plant-ing date, and a PCA was performed on the correlation matrix(Fig. 2). The rst two components accounted for 62% of the totalvariance explained. Consistent with aforementioned results ofPCA on physiochemical parameters and sensory attributes, thesamples separation was driven by a maturity effect on the rstcomponent, while a planting date effect was the main driver ofthe separation on the second component. As shown in the biplot(Fig. 2), peaks 9 and 14 were positively correlated to cucumber ar-oma (r = 0.83 and 0.85), rmness (r = 0.78) and crunchiness(r = 0.78 and 0.77) sensory attributes. In fully ripe fruits, the high-est signicant correlations for fruity aroma were observed withpeak 5 (r = 0.95), peak 2 (r = 0.90) and peak 17 (r = 0.93). Peak 17was also found to be signicantly correlated to sweet taste andfruity avour attributes (r = 0.70 and 0.73).

To evaluate the prediction ability of these correlations, regres-sion analysis was conducted on aforementioned sensory attributesand peaks. Signicant linear relationships were observed betweenfruity aroma and peak 2 (R2 = 0.81; y = 4.8301x 0.5669), peak 5(R2 = 0.78; y = 0.5095646 + 0.0773868x) and peak 17 (R2 = 0.54;

y = 0.3484706 + 0.0919787x), and between peak 17 and fruity a-vour (R2 = 0.70; y = 0.2985612 + 0.1009298x), and sweet taste(R2 = 0.67; y = 0.3883602 + 0.0964181x). Signicant linear relation-ships were also observed between cucumber aroma and peaks 9and 14, although they only explained 38% and 41% of the modelvariation, respectively.

These results indicate that several peaks detected by UFGCSAW may represent markers for these sensory attributes.

This technology has been used in several studies for rapid vola-tile analysis of fresh produce (Du, Olmstead, & Rouseff, 2012; Li,Heinemann, & Irudayaraj, 2007; Li, Wang, Raghavan, & Vigneault,2009; Vallone et al., 2012; Watkins & Wijesundera, 2006). How-ever, to our knowledge, only few studies have focused on fruitmaturity assessment using UFGCSAW.

To evaluate mango (Mangifera indica L.) ripeness, Li et al. (2009)selected a peak, whose concentration increased consecutively therespiration climacteric, and used it to predict mango ripeness,achieving an 80% accuracy rate.

UFGCSAW analysis has also been used in a recent investigationto discriminate among blueberries harvested at three maturitystages (Du et al., 2012). Using 14 selected peaks, a clear separationbetween fully ripe and less ripe berries was successfully achievedfor one of the two cultivars assessed (Primadonna and Jewel).As the authors suggested, the greater separation of the fully ripeberries of the cultivar Primadonna might be ascribed to the high-er level of total volatiles produced by this cultivar.

analy

S. Vallone et al. / Food Chemistry 139 (2013) 171183 181Fig. 2. Biplot from principal component analysis of peaks from UFGCSAW

square = Thunderbird; triangle = Mas Rico; circle = Navigator. Maturity is represented bylight grey = pre-slip dark; lled symbols, dark grey = slip A; lled symbols, black = slip B;axis to a particular parameter indicates how well the variation in that parameter is expsis and sensory attributes. Symbol shape corresponds to different cultivars:

different shades of grey: open symbols, light grey = pre-slip light; lled symbols,open symbols, black = slip C. The length of the dotted arrows from the origin of thelained by the model.

182 S. Vallone et al. / Food Chemistry 139 (2013) 1711834. Conclusions

Maturity at harvest, cultivar and planting date qualitatively andquantitatively affected chemical composition and physical charac-teristics of melon, ultimately impacting sensory perception.

Overall, the perception of sweetness, fruity and musky noteswas greater in ripe fruit, while cucumber notes were predominantin less ripe fruit. Instrumental measurements of esh rmnesswere generally correlated with sensory texture perception. Afteresters, aldehydes were the most abundant group of volatiles inpre-slip fruit. Hexanal, typically imparting green notes to fruit aro-mas, was the predominant aldehyde measured, while 2-heptenalwas signicantly correlated with cucumber aroma in pre-slip fruit.In riper fruit, esters and sulphur containing compounds predomi-nated and correlated with perceived fruity and musky aromas.The cultivar Navigator was generally perceived to have the highestintensity of colour, aroma, taste, and avour attributes, while MasRico scored higher for cucumber notes and rmness. Identifying allthe variables affecting fruit quality and monitoring them duringthe pre- and post-harvest life of a fruit remains a challenge dueto the lack of a single rapid, comprehensive method for physio-chemical analysis.

Volatiles play an important role in determining melon avour,and the signicant correlations existing among various volatilesand sensory attributes found in our and previous studies, suggestthat this relationship might be potentially exploited to predict sen-sory perception by measuring volatiles.

Here, a portable volatile sensing system (UFGCSAW) was ableto discriminate melons of different maturity stages based on 6measured peaks, whose abundances were positively correlated tosensory attributes associated with different maturity stages of mel-on. Our ndings suggest that this technology shows much promisefor future applications in rapid aromameasurement. Further inves-tigations are needed to conrm the ability of the UFGCSAWinstrument in predicting fruit sensory properties.

Acknowledgements

This project was supported by the Specialty Crops Research Ini-tiative Competitive Grants Program Grant No. 2009-51181-05783from the USDA National Institute of Food and Agriculture. We aregrateful to Bill Copes (HM. Clause) for providing the melons usedin this study, and to Robin Clery (Givaudan) for his kind gift of2,3-butanediol diacetate. We thank Edward Orgon, Michael Mace,Aly Depsky and Sharon Wei for technical assistance.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foodchem.2012.12.042.

References

Ayub, R., Guis, M., BenAmor, M., Gillot, L., Roustan, J. P., Latche, A., et al. (1996).Expression of ACC oxidase antisense gene inhibits ripening of cantaloupe melonfruits. Nature Biotechnology, 14(7), 862866.

Bauchot, A. D., Mottram, D. S., Dodson, A. T., & John, P. (1998). Effect ofaminocyclopropane-1-carboxylic acid oxidase antisense gene on theformation of volatile esters in cantaloupe Charentais melon (Cv. Vedrandais).Journal of Agricultural and Food Chemistry, 46(11), 47874792.

Beaulieu, J. C. (2005). Within-season volatile and quality differences in stored fresh-cut cantaloupe cultivars. Journal of Agricultural and Food Chemistry, 53(22),86798687.

Beaulieu, J. C. (2006). Volatile changes in cantaloupe during growth, maturation,and in stored fresh-cuts prepared from fruit harvested at various maturities.

Journal of the American Society for Horticultural Science, 131(1), 127139.Beaulieu, J. C., & Grimm, C. C. (2001). Identication of volatile compounds incantaloupe at various developmental stages using solid phase microextraction.Journal of Agricultural and Food Chemistry, 49(3), 13451352.

Beaulieu, J. C., Ingram, D. A., Lea, J. M., & Bett-Garber, K. L. (2004). Effect of harvestmaturity on the sensory characteristics fresh-cut cantaloupe. Journal of FoodScience, 69(7), S250S258.

Beaulieu, J. C., & Lancaster, V. A. (2007). Correlating volatile compounds, sensoryattributes, and quality parameters in stored fresh-cut cantaloupe. Journal ofAgricultural and Food Chemistry, 55(23), 95039513.

Beaulieu, J. C., & Lea, J. M. (2007). Quality changes in cantaloupe during growth,maturation, and in stored fresh-cut cubes prepared from fruit harvested atvarious maturities. Journal of the American Society for Horticultural Science,132(5), 720728.

Beaulieu, J. C., Lea, J. M., Eggleston, G., & Peralta-Inga, Z. (2003). Sugar and organicacid variations in commercial cantaloupes and their inbred parents. Journal ofthe American Society for Horticultural Science, 128(4), 531536.

Borgognone, M., Bussi, J., & Hough, G. (2001). Principal component analysis insensory analysis: Covariance or correlation matrix? Food Quality and Preference,12, 323326.

Chrost, B., & Schmitz, K. (1997). Changes in soluble sugar and activity of alpha-galactosidases and acid invertase during muskmelon (Cucumis melo L) fruitdevelopment. Journal of Plant Physiology, 151(1), 4150.

Dai, N., Cohen, S., Portnoy, V., Tzuri, G., Harel-Beja, R., Pompan-Lotan, M., et al.(2011). Metabolism of soluble sugars in developing melon fruit: A globaltranscriptional view of the metabolic transition to sucrose accumulation. PlantMolecular Biology, 76(12), 118.

Du, X., Olmstead, J., & Rouseff, R. (2012). Comparison of fast gas chomatographysurface acoustic wave (FGCSAW) detection and GCMS for characterizingblueberry cultivars and maturity. Journal of Agricultural and Food Chemistry, 60,50995106.

El-Sharkawy, I., Manriquez, D., Flores, F. B., Regad, F., Bouzayen, M., Latche, A., et al.(2005). Functional characterization of a melon alcohol acyl-transferase genefamily involved in the biosynthesis of ester volatiles. Identication of the crucialrole of a threonine residue for enzyme activity. Plant Molecular Biology, 59(2),345362.

Flores, F., El Yahyaoui, F., de Billerbeck, G., Romojaro, F., Latche, A., Bouzayen, M.,et al. (2002). Role of ethylene in the biosynthetic pathway of aliphatic esteraroma volatiles in Charentais cantaloupe melons. Journal of Experimental Botany,53(367), 201206.

Fuke, S., & Ueda, Y. (1996). Interactions between umami and other avorcharacteristics. Trends in Food Science & Technology, 7(12), 407411 [Article].

Grosch, W., & Schwarz, J. M. (1971). Linoleic and linolenic acid as precursors of thecucumber avor. Lipids, 6, 351352.

Hadeld, K. A., Dang, T., Guis, M., Pech, J. C., Bouzayen, M., & Bennett, A. B. (2000).Characterization of ripening-regulated cDNAs and their expression in ethylene-suppressed Charentais melon fruit. Plant Physiology, 122(3), 977983.

Harrell, F. E. and with contributions from many other users (2012). Hmisc: HarrellMiscellaneous. R package version 3.9-1. .

Heyer, B. R., Taylor-Burds, C. C., Tran, L. H., & Delay, E. R. (2003). Monosodiumglutamate and sweet taste: Generalization of conditioned taste aversionbetween glutamate and sweet stimuli in rats. Chemical Senses, 28(7), 631641.

Homatidou, V. I., Karvouni, S. S., Dourtoglou, V. G., & Poulos, C. N. (1992).Determination of total volatile components of Cucumis melo L. varietycantaloupensis. Journal of Agricultural and Food Chemistry, 40(8), 13851388.

Jordan, M. J., Shaw, P. E., & Goodner, K. L. (2001). Volatile components in aqueousessence and fresh fruit of Cucumis melo cv. athena (muskmelon) by GCMS andGCO. Journal of Agricultural and Food Chemistry, 49(12), 59295933.

Kader, A. A. (2008). Perspective: Flavor quality of fruits and vegetables. Journal of theScience of Food and Agriculture, 88, 18631868.

Kemp, S. E., & Beauchamp, G. K. (1994). Flavor modication by sodium chloride andmonosodium glutamate. Journal of Food Science, 59(3), 682686.

Kourkoutas, D., Elmore, J. S., & Mottram, D. S. (2006). Comparison of the volatilecompositions and avour properties of cantaloupe, galia and honeydewmuskmelons. Food Chemistry, 97(1), 95102.

Lester, G. (2006). Consumer preference quality attributes of melon fruits. ActaHorticulture, 712, 175182.

Li, C., Heinemann, P. H., & Irudayaraj, J. (2007). Detection of apple deteriorationusing an electronic nose and zNose (TM). Transactions of the Asabe, 50(4),14171425.

Li, X. D., Staszewski, L., Xu, H., Durick, K., Zoller, M., & Adler, E. (2002). Humanreceptors for sweet and umami taste. Proceedings of the National academy ofSciences of the United States of America, 99(7), 46924696.

Li, Z. F., Wang, N., Raghavan, G. S. V., & Vigneault, C. (2009). Ripeness and rotevaluation of Tommy Atkins mango fruit through volatiles detection. Journal ofFood Engineering, 91(2), 319324.

Maga, J. A. (1983). Flavor potentiators. Crc Critical Reviews in Food Science andNutrition, 18(3), 231312.

Mizuno, T., Kato, K., Harada, M., Miyajima, Y., & Suzuki, E. (1971). Studies on the freesugars and amino acids in a fruit of muskmelon. Journal of The Japanese Societyfor Food Science and Technology, 18, 319325.

Ninomiya, Y., Nakashima, K., Fukuda, A., Nishino, H., Sugimura, T., Hino, A., et al.

(2000). Responses to umami substances in taste bud cells innervated by the

chorda tympani and glossopharyngeal nerves. Journal of Nutrition, 130(4),950S953S.

Nishiyama, K., Guis, M., Rose, J. K. C., Kubo, Y., Bennett, K. A., Lu, W. J., et al. (2007).Ethylene regulation of fruit softening and cell wall disassembly in Charentaismelon. Journal of Experimental Botany, 58(6), 12811290.

Obando-Ulloa, J. M., Ruiz, J., Monforte, A. J., & Fernandez-Trujillo, J. P. (2010). Aromaprole of a collection of near-isogenic lines of melon (Cucumis melo L.). FoodChemistry, 118(3), 815822.

Pech, J. C., Bouzayen, M., & Latche, A. (2008). Climacteric fruit ripening: Ethylene-dependent and independent regulation of ripening pathways in melon fruit.Plant Science, 175(12), 114120.

Pratt, H. (1971). Melons. The biochemistry of fruits and their products. NY: AcadamicPress. 207232.

Sako, N., & Yamamoto, T. (1999). Analyses of taste nerve responses with specialreference to possible receptor mechanisms of umami taste in the rat.Neuroscience Letters, 261(12), 109112.

Schieberle, P., Ofner, S., & Grosch, W. (1990). Evaluation of potent odorants incucumbers (Cucumis sativus) and muskmelons (Cucumis melo) by aroma extractdilution analysis. Journal of Food Science, 55(1), 193195.

Senesi, E., Di Cesare, L. F., Prinzivalli, C., & Lo Scalzo, R. (2005). Inuence of ripeningstage on volatiles composition, physicochemical indexes and sensoryevaluation in two varieties of muskmelon (Cucumis melo L var reticulatusNaud). Journal of the Science of Food and Agriculture, 85(8), 12411251.

Senesi, E., Lo Scalzo, R., Prinzivalli, C., & Testoni, A. (2002). Relationships betweenvolatile composition and sensory evaluation in eight varieties of nettedmuskmelon (Cucumis melo L var reticulatus Naud). Journal of the Science ofFood and Agriculture, 82(6), 655662.

Stein, S. E. (1999). An integrated method for spectrum extraction and compoundidentication from gas chromatography/mass spectrometry data. Journal of theAmerican Society for Mass Spectrometry, 10(8), 770781.

Stone, H., Sidel, J., Oliver, S., Woolsey, A., & Singleton, R. C. (1974). Sensoryevaluation by quantitative descriptive analysis. Food Technology, 28(11), 2434.

Thimijan, R. W., & Heins, R. D. (1983). Photometric, radiometric, and quantum lightunits of measure A review of procedures for interconversion. HortScience,18(6), 818822.

Tomic, O., Luciano, G., Nilsen, A., Hyldig, G., Lorensen, K., & Naes, T. (2010).Analysing sensory panel performance in a prociency test using the PanelChecksoftware. European Food Research and Technology, 230(3), 497511.

Ullrich, F., & Grosch, W. (1987). Identication of the most intense volatile avor

USDA-ERS. (2010, 05/2011). Vegetables and melons yearbook. Accessed 6.1.2012.

Vallone, S., Lloyd, N. W., Ebeler, S. E., & Zakharov, F. (2012). Fruit volatile analysisusing an electronic nose. Journal of Visualized Experiments (61), e3821.

Vermeir, S., Nicolai, B. M., Jans, K., Maes, G., & Lammertyn, J. (2007). High-throughput microplate enzymatic assays for fast sugar and acid quanticationin apple and tomato. Journal of Agricultural and Food Chemistry, 55(9),32403248.

Wang, Y. M., Wyllie, S. G., & Leach, D. N. (1996). Chemical changes during thedevelopment and ripening of the fruit of Cucumis melo (cv Makdimon). Journalof Agricultural and Food Chemistry, 44(1), 210216.

Watkins, P., & Wijesundera, C. (2006). Application of zNose (TM) for the analysis ofselected grape aroma compounds. Talanta, 70(3), 595601.

Wyllie, S. G., Leach, D. N., Wang, Y. M., & Shewfelt, R. L. (1995). Key aromacompounds in melons Their development and cultivar dependence fruit avors:Biogenesis, characterization, and authentication (Vol. 596, pp. 248257).Washington, DC: American Chemical Society.

Wyllie, S. G., Leach, D. N., & Wang, Y. (1996). Development of avor attributes in thefruit of C. melo during ripening and storage. In G. R. Takeoka, R. Teranishi, P. J.Williams, & A. Kobayashi (Eds.), Biotechnology for improved foods and avor(pp. 228239). Washington, DC: American Chemical Society.

Wyllie, S. G., & Leach, D. N. (1990). Aroma volatiles of Cucumis melo cv GoldenCrispy. Journal of Agricultural and Food Chemistry, 38(11), 20422044.

Wyllie, S. G., & Leach, D. N. (1992). Sulfur-containing compounds in the aromavolatiles of melons (Cucumis melo). Journal of Agricultural and Food Chemistry,40(2), 253256.

Wyllie, S. G., Leach, D. N., Wang, Y. M., & Shewfelt, R. L. (1994). Sulfur volatiles inCucumis melo Cv Makdimon (muskmelon) aroma Sensory evaluation by gas-chromatography olfactometry. In C. J. Mussinan & M. E. Keelan (Eds.). Sulfurcompounds in foods (Vol. 564, pp. 3648). Washington: American ChemicalSociety.

Yamaguchi, M., Hughes, D. L., Yabumoto, K., & Jennings, W. G. (1977). Quality ofcantaloupe muskmelons Variability and attributes. Scientia Horticulturae, 6(1),5970.

Yamamoto, T., Matsuo, R., Fujimoto, Y., Fukunaga, I., Miyasaka, A., & Imoto, T.(1991). Electrophysiological and behavioral studies on the taste of umamisubstances in the rat. Physiology & Behavior, 49(5), 919925.

Yeomans, M. R., Gould, N. J., Mobini, S., & Prescott, J. (2008). Acquired avoracceptance and intake facilitated by monosodium glutamate in humans.

S. Vallone et al. / Food Chemistry 139 (2013) 171183 183compounds formed during autoxidation of linoleic acid. Zeitschrift FurLebensmittel-Untersuchung Und-Forschung, 184(4), 277282.Physiology & Behavior, 93(45), 958966.